The new build 2322 doesn’t seem to fix it, though it might save just a slight bit faster. The reason this has me concerned is that anything I experience on my computer could easily be significantly amplified on a server with 20 people on it. Though on a hopeful side, our server has SSD, which I can’t say the same for my computer. I just have a fast magnetic HDD.
I figured out that stargate keeps chunks loaded for each stargate. I compared chunks loaded with and without stargate on. The difference is just above the number of stargates, which I imagine is due to some stargates overlapping chunk boundaries. I’m not sure where in the code this is happening. Maybe they all “open” at the start and then it never unloads them?
Remember how I said /save-all didn’t lag it? Well, it didn’t occur to me that this particular 45-second interval was tied to save-on/off. (Save-off was confusing in the past. At a certain point I couldn’t tell the difference between it on/off, and there was no way to return its state. There was also a bug that made the feature not work for a while.) Someone in the bug report reminded me of that. I’ve used save-off in the past. It was necessary over a year ago to reduce lag. So I tried it here and lo and behold… it stopped the lag. I’m already doing a save-all interval on our server. If I make it /save-on \ /save-all \ /save-off then it won’t lag, and it frees up the memory that save-off builds up.
I think with this new information I’m comfortable with going ahead with updating late tonight. I still think Drakia needs to look into this issue for the general public, even if it’s not his fault.
Oh, can you look at /save-on /save-off and /save-all? I’m not exactly sure what’s going on here. I’m trying to figure out what each of these actually do. worldserver.canSave makes no sense to me. Here’s a couple key sections: